Lime Dissolution in Foaming BOF Slag · 2018-06-25 · Lime Dissolution in Foaming BOF Slag JOHAN MARTINSSON, BJO¨RN GLASER, and DU SICHEN The paper describes the dissolution mechanisms

Lime Dissolution in Foaming BOF Slag

JOHAN MARTINSSON, BJORN GLASER, and DU SICHEN

The paper describes the dissolution mechanisms of lime into liquid and foaming slags relevant tothe BOF process. Two different master slags are employed, representing two different periods ofthe converter process: an early stage where the FeO content is fixed to 45 wt pct, and a laterstage where the FeO content is fixed to 25 wt pct. For these master slags, the ratio between CaO/SiO2 is varied to examine the effect of basicity on lime dissolution. Calcium silicates are formedand peeled off, or partially peeled off, from the interface between the lime cube and the slag in allcases. The main difference for the dissolutions in pure liquid slag and foaming slag is thecontrolling step for dissolution. In liquid slag, the controlling mechanism is the removal of thecalcium silicate layers, while in foaming slag, the controlling mechanism is the contact areabetween the lime and the liquid slag phase of the foam. The strong convection in the foamenhance the dissolution process, in some cases, the lime even dissociates into small pieces.

https://doi.org/10.1007/s11663-018-1421-6� The Author(s) 2018

I. INTRODUCTION

THE dissolution of lime into BOF slags has been thetopic for several researchers and is of great interest forthe steel industry.[1–14] The dissolution of lime andconsequently the increase of slag basicity enhances thedephosphorization process. An increase in basicity isalso important to obtain a lower viscosity of the slag. Avery acidic slag would increase the size of the silica ionsand increase the viscosity, which in turn has shown todecrease the foaming ability of the slag.[15,16]

It is well documented that the CaO content increasesrapidly in the beginning of the BOF process andincreases slowly after approximately 5 minutes of theblowing time.[17] It is also described that foam starts toform profoundly after approximately 3 to 5 minutes ofthe process. The industry is interested in good data ofthe lime dissolution to develop/improve their processmodels for better process control.

Evans et al. and Deng et al. studied the dissolutionmechanisms of lime in different liquid BOF slagscontaining CaO-FeO-SiO2.

[1,7–10] It was found that thelime reacts with SiO2 in the liquid slag, forming solidcalcium silicate layers, 2CaOÆSiO2, in the interfacebetween lime and slag. The controlling dissolutionmechanism is the removal of these interfacial layers byshear stress, irrespective of the initial CaO content in the

slag. It was also found that slag with higher FeO contentdissolves the CaO faster. The FeO penetrates into thelime and helps the detachment of the calcium silicatelayers. On the other hand, when the FeO content is low,the calcium silicate layer is dense. The dense layerprotects the lime from slag penetration and thereforedecreases the dissolution rate.[7] Maruoka et al. studiedthe possibility of using quicklime where a core of CaCO3

remained. It was suggested that the decompositionwould enhance the dissolution process since the gener-ated CO2 gas would help to carry away the calciumsilicate layer.[14]

To the best knowledge of the present authors, nolaboratory study on lime dissolution in foaming slag hasbeen reported. By addition of carbon saturated iron intoa FeO containing slag, a massive gas evolution isgenerated. The aim of this work is to find out if there is adifference between lime dissolution in foaming slagcompared to its dissolution in pure liquid slag.

II. EXPERIMENTS

The slag evolution in the BOF process can be dividedinto two main periods, viz. an early stage when the FeOcontent is above 40 wt pct, and a later stage when theFeO content has decreased to 20 to 25 wt pct due to thedecarburization.[17] In view of this change in slagcomposition, two different master slags were studied inthis paper, one representing the early period where theFeO content was fixed at 45 wt pct, and a later stagewhere the FeO content was fixed at 25 wt pct. The CaOcontent was chosen to vary from 0 to 30 wt pct in bothmaster slags to evaluate how the lime dissolution ratevaried at different initial CaO content. The slag

JOHAN MARTINSSON, BJORN GLASER, and DU SICHENare with the Department of Materials Science and Engineering, KTHRoyal Institute of Technology, 100 44 Stockholm, Sweden. Contacte-mail: [email protected]

Manuscript submitted June 25, 2018.Article published online October 12, 2018.

3164—VOLUME 49B, DECEMBER 2018 METALLURGICAL AND MATERIALS TRANSACTIONS B

compositions are listed in Table I. All slags were utilizedin the liquid slag experiments, while only slags A, B, C,and D were chosen for the foaming slag experiments,since the foaming in the BOF process starts appreciablyafter lime has started to dissolve into the slag. Also thetheoretical density and dynamic viscosity of the slags areincluded to the table.[18] The density, qtheoretic, is given in(kg/m3) and viscosity, ldynamic, in (Pa s). The viscositydata are based on iso-lines from the reference and aretherefore given in ranges.[18]

A. Material Preparations

The FeO was produced by mixing Fe2O3 and ironpowder with totally 51 at. pct oxygen. The mixture waskept in a closed iron crucible at 1143 K for 60 hours inargon atmosphere. The sintered body was then crushedinto small pieces. The formation of FeO was confirmedby XRD analysis.

The CaO powder was calcined at 1173 K for 10 hours.SiO2 and MnO were dried at 383 K for approximately24 hours.

Burnt lime was provided by a supplier to TATA Steel,and no further calcination was carried out before theexperiments. The lime was cut and grinded into 1.0 91.0 9 1.0 cm cubes. The quality of the lime varied andimpurities were detected. The density of the cubes wasfound to vary between approximately 1700 and 2500 kg/m3. Cubes with the density of approximately 2000 kg/m3

were chosen for the experiments to minimize the qualityvariation.

B. Dissolution of Lime in Liquid Slag

A vertical tube furnace was employed for the exper-iments of lime dissolution in liquid slag. The setup isschematically described in Figure 1. An alumina tubewas used as reaction chamber. The furnace wasequipped with super Kanthal heating elements. A typeB thermocouple was used to control the furnacetemperature. Another thermocouple of type B wasinserted from the bottom of the reaction tube and keptjust beneath the sample to record the sampletemperature.

The slag components were mixed thoroughly intoselected composition. In the case of slags 0, A, B, and C,an amount of 30 g mixed powder was put into amolybdenum crucible with an inner height of 45 mmand inner diameter of 30 mm. In the case of slag D, only20 g of slag could be put in the Mo crucible (due to itslow density). The small molybdenum crucibles were thenplaced in a molybdenum holder which was connected toa steel tube and a lifting system. A molybdenum impellerwas inserted inside the steel tube and connected to astirring motor placed on top of the lifting system. Withthis arrangement, the slag bath along with the CaOsample could be stirred during the experiments.

The slags were pre-melted for one hour and thenquenched to assure that the slags were homogeneousbefore the lime cubes were introduced into the slag. Alime cube was put on top of the solidified slag and was

then lowered down to a resting position of the reactiontube and kept there for 15 minutes. The temperature ofthe resting position was just above the melting temper-ature of the slag. The sample was then lowered downfast to the heating zone at which point the entire slagmelted at the same time. The purpose of using theresting position was to ensure a homogeneous temper-ature in the slag which allows a better control of themelting moment of the slag. This moment could easilybe controlled by lowering down the stirring impeller andtouch the surface of the slag. The stirring startedimmediately when the slag was molten and was ongoingfor 30, 60, or 180 seconds. The stirrer was then raised upfrom the slag after which the sample was quenched inthe cooling chamber.

C. Dissolution of Lime in Foaming Slag

An induction furnace with a water-cooled copper coilwas employed to generate the foams, see the experi-mental setup in Figure 2. The furnace was set to 270 V,25 A, and 26 kHz during the experiments.1 g of graphite powder and 7 g of pig iron containing

3.9 wt pct carbon were put in the bottom of a graphitecrucible with inner height of 140 mm, inner diameter of30 mm, and a wall thickness of 10 mm. 67 g of a selectedslag was then added on the top of the hot metal. Thecrucible was painted with Yttrium Oxide paint to avoidheavy oxidation of the graphite during the experiment.The temperature was measured using a calibratedinfrared temperature sensor of model thermoMETERCTM-1SF75-C3, on consideration that regular thermo-couples would be affected by the magnetic field andcould therefore not be used.[19] The slag was moltenapproximately 3 minutes after the furnace was turned onand the foaming started a couple of seconds later duethe reduction of FeO by the carbon in the hot metal.When the foam was well developed and an appropriatetemperature was reach at around 1873 K, a lime cubewas dropped into the foam from the top, and a stopwatch was started. The furnace was turned off after 30,60, or 120 seconds and the slag was solidified sponta-neously within seconds.

D. Sample Preparation and Examination

Both the samples along with the crucibles from liquidslag and foaming slag experiments were cut horizontallyafter the experiments. The size of the remaining limecubes were measured with a pair of calipers. Thesamples were then mounted to pellets in epoxy. Thepellets were grinded and polished in ethanol beforeexamination in a scanning electron microscope (SEM).SEM was employed for the examination of the

interface between lime cube and slag. The pictures weretaken with backscattered electrons (BSE). Energy-dis-persive X-ray spectroscopy (EDS) was employed forpoint analysis and map analysis. This information wasessential for an understanding of the dissolution mech-anisms. Many pores and cracks were detected inside thelime cube. They appear during sample preparation due

METALLURGICAL AND MATERIALS TRANSACTIONS B VOLUME 49B, DECEMBER 2018—3165

to the poor strength of the lime, and they are easilydetected as black areas in the SEM pictures (marked inthe figures discussed later).

III. RESULTS

A. Lime Dissolution in Liquid Slag

Figure 3 presents the remaining cube sizes of the limefrom the dissolution experiments in liquid slag. Asexpected, the lime dissolution is fastest in slag 0, wherethe initial CaO content is zero, and therefore has thestrongest driving force for dissolution due to theconcentration gradient. The cube is dissolved into theslag already after 60 seconds. The figure also shows thatslags A, B, and 0, where the initial FeO content is 45 wtpct, have faster lime dissolution than slags C and D,where the initial FeO content is only 25 wt pct. Thisresult is in good agreement with the literature.[7] The fastdissolution of lime is due to that FeO facilitates theremoval of calcium silicates.

Figures 4(a) through (c) illustrate the dissolutionprocess of a lime cube into slag B. The penetration ofslag into the cube is also well brought out byFigure 4(a).The SEM microphotograph of the interface between

lime cube and liquid slag B after 60 seconds stirring ispresented in Figure 5. Figures 4 and 5 are in very goodaccordance with the results of previous works by Evanset al. and Deng et al.[1,7,9] As shown in Figure 5, a layerof 2CaOÆSiO2 about 20 to 30 lm in thickness has beenformed in the interface between slag B and lime cubeafter 60 seconds. It should be mentioned that a layer of2CaOÆSiO2 is detected in all samples after the stirring.On the other hand, the thickness depends on thecomposition of the slag and the reaction time. Figure 5also shows how islands of calcium silicates are torn offfrom the cube surface and brought into the slag phase.The fact that the main population of the small calciumsilicate islands are very close to the surface of the cubeimplies that they are flushed off from the surface of thecube due to stirring. The distance between the smallislands and the layer of calcium silicates reveals that

Table I. Slag Compositions in Weight Percent

Slag A B 0 C D

FeO 45 45 45 25 25CaO 30 15 0 30 15SiO2 25 40 40 25 40MnO 0 0 15 20 20qtheoretic 3500 3300 3500 3400 3100ldynamic 0.05 to 0.1 0.15 to 0.2 0.16 to 0.24 0.05 to 0.1 0.15 to 0.2

Fig. 1—Experimental setup for liquid slag.

3166—VOLUME 49B, DECEMBER 2018 METALLURGICAL AND MATERIALS TRANSACTIONS B

these islands have already detached from the layer. Onecan also see liquid slag containing mostly FeO inside thelime cube. The depth of pure liquid slag could also beseen varying in the samples. Slags A and B had deeperliquid slag penetration than slags C and D, which is alsoin agreement with the results of the literature.As seen in Figure 5, some small amounts of 3CaOÆ-

SiO2 are formed in the CaO matrix near the interface.The formation of 3CaOÆSiO2 instead of 2CaOÆSiO2 inthis region can be well explained by the high CaOactivity (aCaO=1) in the region. The formation of3CaOÆSiO2 would also consume SiO2 in the liquid,leading to high FeO content in the penetrating liquidphase as mentioned above. The black areas are (asmentioned earlier) pores of missing lime pieces that wastorn off during the sample preparation.

B. Lime Dissolution in Foaming Slag

Figure 6 presents the results of the dissolution of limecubes in foaming slag. The results show that thedissolution rates in foaming slags A and C are similarto the lime dissolution in the corresponding liquid slags,while the dissolution of lime in foaming slags B and Dare faster than in the corresponding liquid slags. The

Fig. 2—Experimental setup for foaming slag.

Fig. 3—Dissolution of lime in liquid slag.

Fig. 4—Dissolution of lime in slag B after (a) 30 s, (b) 60 s, and (c) 120 s.

Fig. 5—Interface between lime cube and liquid slag B after 60 s.

METALLURGICAL AND MATERIALS TRANSACTIONS B VOLUME 49B, DECEMBER 2018—3167

dissolution in foaming slag D differs greatly from thetrend; the lime cube was found dissociated to smallpieces after 120 seconds, see Figure 7. This experimentwas repeated several times and the outcome was alwaysthe same, which will be discussed later.

The SEM picture shown in Figure 8 reveals a discon-tinuous layer of gas phase separating big parts of thelime cube from the liquid phase of the slag. This resultwas obtained with slag C after 120 seconds of reactiontime. The presence of the gas phase layer at the interfacewould reduce the contact area between the slag and thecube.

As the lime dissolves into the slag, an interfacial layerof calcium silicates is found between the cube and thefoaming slag, although the layer is thinner compared tothe case of pure liquid slag. Figure 9 shows that theinterfacial layer of calcium silicate is only approximately5 lm after reaction with the foaming slag B for 60seconds. Similar to the case of pure liquid slag, the liquidslag that has penetrated into the CaO matrix containsmostly FeO. The thickness of the penetrated layer isusually less than 100 lm.

IV. DISCUSSION

The dissolution mechanisms of lime into liquid BOFslags have been well documented by Evans et al. andDeng et al.[1,7–10] These mechanisms have further beenconfirmed by the present results. An example is shown inFigure 5. As mentioned in the result part, calciumsilicates have been formed at the interface between cubeand slag, and islands of calcium silicates can be seenleaving the cube into the slag phase. Liquid slagcontaining FeO, CaO, and small amounts of SiO2 (andsome MnO for slag C and D) can also be seen inside thecube. The formation of 3CaOÆSiO2 has resulted in verylow SiO2 content in the liquid phase among the CaOmatrix.Some studies has shown that these solid calcium

silicates are important for the dephosphorization reac-tion.[20–24] Since the main focus of this work is the limedissolution, no phosphorous was added to the system.However, it would be an interesting and valuablecontinuation to study the mechanisms of dephospho-rization in foaming slag and to examine whether thegenerated gas would also be able to remove the calciumsilicate layer with the presence of phosphorous.The dissolution of lime into liquid slags as functions

of slag composition is presented in Figure 3. The rapiddissolution rate observed in Slag 0 is not surprising sincethe initial CaO content is 0 and thus has the strongestdriving force to dissolve the lime cube. It is also shownthat the dissolution rates are faster in slags A and Bcompared to slags C and D, which confirms the theorythat a high FeO content dissolves the lime faster.[7] Liet al. showed that both Fe2+ and Mn2+ ions facilitatethe inward diffusion of slag into lime, which increases

Fig. 6—Dissolution of lime in foaming slag.

Fig. 7—Dissociated lime cube in foaming slag D after 120 s.

Fig. 8—Upper right corner of lime cube in foaming slag C after 120 s.

Fig. 9—Interface between cube and foaming slag B after 60 s.

3168—VOLUME 49B, DECEMBER 2018 METALLURGICAL AND MATERIALS TRANSACTIONS B

the dissolution rate.[12] The present experiments show afaster dissolution rate when using 45 wt pct FeO insteadof 25 wt pct FeO and 20 wt pct MnO (comparing slag Aand C). The Fe2+ ion is smaller than the Mn2+, andcould (along with a less complex slag) perhaps facilitatethe inward diffusion, and hence increase the dissolutionrate. However, more detailed studies are needed before areliable conclusion can be made. Industrial experiencesusing lime particles coated with FeO and/or MnO hasalso shown that the coating substantially helps the limedissolution. However, the high cost of the coated limehas hindered the development of this application. Infact, the fast dissolution of FeO-coated lime is inaccordance with earlier laboratory observationsobtained in the present lab (Prof. Du Sichen, Depart-ment of Materials Science and Engineering, RoyalInstitute of Technology, Stockholm, Sweden, 2018,Private communication). Nevertheless, further studiesare strongly recommended to find clarity in the matter.

Comparing the results to the work by Maruoka et al.,one major difference can be seen. Maruoka et al. foundtwo periods of gas generation from the quicklime duringthe initial stage after exposure to the slag.[14] CO2 wasgenerated by thermal decomposition of CaCO3, andcontinued for totally 150 seconds. Yet in this work,Figure 4 evidently shows that no gas was generatedduring the experiments in pure liquid slag. The absenceof gas in these experiments could be due to the type oflime which might be different compared to the workused by Maruoka et al. The decomposition of CaCO3

was also studied by Deng et al. It is worthwhile tomention that the inner part of the sample did not reachthe decomposition temperature, even after 180 seconds,when the lime piece was put into a slag at 1823 K.[10]

Note that the main focus of the present work is tostudy the difference in lime dissolution mechanisms infoaming slag in comparison with the same in pure liquidslag.

Three main differences have been observed betweenthe dissolution of lime in foaming and liquid slag. In thefoaming slag, it is evident that the gas phase interruptsthe dissolution reaction of lime as indicated by thefollowing observation:

1. In Figure 8, gas is present at the interface betweenthe cube and the slag. The gas diminishes the contactarea between the lime and the liquid phase of theslag.

2. The thickness of the calcium silicate layers at the cubeinterface are shown to be different between liquid andfoaming slag. In liquid slag, the layer thickness inslag B is approximately 30 lm after 60 seconds, whileit is only approximately 5 lm in correspondingfoaming slag. This aspect is clearly brought out by acomparison of Figures 5 and 9.

3. Figures 5 and 9 also show that the depth of slagpenetration is smaller in the foaming slag comparedto liquid slag, only 100 lm instead of 200 lm.

The presence of a discontinue layer of gas between thecube and slag in the case of foaming experiment couldbe due to the gas accumulation. The higher friction tothe gas movement at the solid surface could be the

reason for the formation of this layer. However, thisaspect would need a careful and systematic study in thefuture. Nevertheless, the presence of gas phase betweenthe sample and foaming slag reduces the area ofreaction.The thinner layer of calcium silicate in the case of

foaming slag could be explained by the chaotic envi-ronment that occurs in the foaming slags. The violentmovement of the foam would push the lime cube up anddown, which would enhance the detachment of thecalcium silicate pieces. In addition, the busting of gasfrom the cube surface would further help the flushing offof the calcium silicate pieces. It is reasonable to expectthat the thinner calcium layer will enhance the dissolu-tion process.The smaller depth of slag penetration in the foam in

comparison with the case of liquid slag would be a directresult of the presence of gas layer and less contact areabetween the slag and the solid sample. In fact, the lessslag penetration is expected to result in lower dissolutionof lime. A combination of the above-mentioned threedifferences would lead to different situations for thedifferent slag compositions and their behavior.For slag compositions A and C, the dissolution rate is

very similar in the foaming slag and liquid slag.Although the reaction might be slightly faster in liquidslag, the removal of calcium silicate layers is faster in thefoaming slag. All in all, the dissolution mechanismsseem to even out and in total, the dissolution rates turnout to be the same for both systems. Slag A and slag Cboth have high basicity and relatively low viscosities. Itwas found that these slags generate uniform foams in anearlier work.[15] Because of the violent movement of thefoam, as discussed earlier, (1) the faster removal ofcalcium silicate pieces and (2) the reduce area of theinterface seem to compensate each other. The joint effectof the two factors leads to a dissolution rate very similarto the rate in pure liquid slag.Slag B has high viscosity. It was found that slags with

too high viscosity does not generate uniform foams.[15]

The gas phase does not exist as small gas bubbles, but aslong gas channels. The cross section of a typical samplecontaining a gas channel is shown in Figure 10.

Fig. 10—Cross section of sample containing gas channel.

METALLURGICAL AND MATERIALS TRANSACTIONS B VOLUME 49B, DECEMBER 2018—3169

Although some smaller bubbles can also be observed inthe figure, the majority of the gas joins the channel witha diameter of approximately 10 mm. The gas channelscreate a chaotic system where the slag surface oscillatesrapidly up and down as the big gas channels rise towardthe surface and punctuate the surface allowing the gas toescape. This violent movement of the slag phase wouldgreatly enhance the removal of the calcium silicate layerat the surface of the lime surface. This would explainwell the faster dissolution of lime in this ‘‘foaming’’ slagcompared to the dissolution in the pure liquid slag.

The chaotic behavior that was generated in slag B wasalso seen in the case of slag D. However, in ‘‘foaming’’slag D, the lime dissociated into small pieces after 60seconds into the experiments. The experiment wasrepeated several times and the outcome was always thesame. The violent movement of the slag due to theformation of gas channels would be responsible for thelime dissociation. The dissociation of CaO generateslarger reaction area between liquid slag and the lime.The bigger contact area enhances greatly the dissolutionprocess. This would explain the profound increase of thedissolution of slag D in ‘‘foam’’ slag in comparison withthe case of pure slag.

Hence, while the controlling dissolution mechanism inliquid slag is the removal of interfacial layers, thecontrolling dissolution mechanism in foaming slag is thecontact area between the lime and liquid phase of thefoaming slag.

The dissolution of lime in the later stage of the BOFprocess is best represented by the experiments conductedin foaming slag C. During this period, as revealed by thepresent experimental result, the lime dissolution is veryslow, almost no dissolution during two minutes. Notethat the CaO content shows a certain increase duringthis period according to the literature. However, thisincrease could be mostly due to the decrease of FeO bydecarburization. If the amount of FeO decreases, thecontent of CaO will consequently increase.

The experimental results that might be the mostinteresting for the lime dissolution process in the BOFfurnace is the dissolution of lime cubes in the liquid slag0 and in the foaming slag B. While the dissolution oflime in pure Slag 0 is relevant to the initial stage of theBOF process (little foaming of slag is present), thedissolution of lime in both liquid slag B and foamingslag B would through some lights on the behavior oflime in the slag a few minutes after the start of theblowing.

V. SUMMARY

Lime cubes were dropped into liquid and foamingslags relevant to BOF process. It was found that thedissolution rate was very similar in slags (A and C)having lower dynamic viscosities. The dissolution infoaming slags B and D, both of which had higherdynamic viscosities, was faster than in the same liquidslags. Lime in foaming slag D dissociated to small piecesafter 60 seconds of experiment, which led to even faster

dissolution. While the controlling dissolution mecha-nism in liquid slag was the removal of interfacial layer,the controlling dissolution mechanism in foaming slagwas the contact area between the lime and liquid phaseof the foaming slag.

OPEN ACCESS

This article is distributed under the terms of theCreative Commons Attribution 4.0 International Li-cense (http://creativecommons.org/licenses/by/4.0/),which permits unrestricted use, distribution, andreproduction in any medium, provided you giveappropriate credit to the original author(s) and thesource, provide a link to the Creative Commons li-cense, and indicate if changes were made.

REFERENCES1. C.A. Natalie and J.W. Ewans: Ironmak. Steelmak., 1979, vol. 6,

pp. 101–09.2. T. Hamano, M. Horibe, and K. Ito: ISIJ Int., 2004, vol. 44,

pp. 263–67.3. T. Hamano, S. Fukagai, and F. Tsukihashi: ISIJ Int., 2006,

vol. 46, pp. 490–95.4. S.H. Amini, M.P. Brungs, O. Ostrovski, and S. Jahanshani: Me-

tall. Mater. Trans. B, 2006, vol. 37B, pp. 773–80.5. S. Amini, M. Brungs, and O. Ostrovski: ISIJ Int., 2007, vol. 47,

pp. 32–37.6. J. Yang, M. Kuwabara, T. Asano, A. Chuma, and J. Du: ISIJ Int.,

2007, vol. 47, pp. 1401–08.7. T. Deng, J. Gran, and D. Sichen: Steel Res. Int., 2010, vol. 81,

pp. 347–55.8. T. Deng, B. Glaser, and D. Sichen: Steel Res. Int., 2012, vol. 83,

pp. 259–68.9. T. Deng and D. Sichen: Metall. Mater. Trans. B, 2012, vol. 43B,

pp. 578–86.10. T. Deng, P. Nortier, M. Ek, and D. Sichen: Metall. Mater. Trans.

B, 2013, vol. 44B, pp. 98–105.11. N. Maruoka, A. Ishikawa, H. Shibata, and S.Y. Kitamura: High

Temp. Mater. Process., 2013, vol. 32, pp. 15–24.12. Z.S. Li, M. Whitwood, S. Millman, and J. van Boggelen: Ironmak.

Steelmak., 2014, vol. 41, pp. 112–20.13. X. Guo, Z.H.I. Sun, J. Van Dyck, M. Guo, and B. Blanpain: Ind.

Eng. Chem. Res., 2014, vol. 53, pp. 6325–33.14. N. Maruoka, A. Ito, M. Hayasaka, S.Y. Kitamura, and H. No-

gami: ISIJ Int., 2017, vol. 57, pp. 1677–83.15. J. Martinsson, B. Glaser and D. Sichen: Ironmak. Steelmak., 2017,

pp. 1–5.16. L.S. Wu, G.J. Albertsson, and D. Sichen: Ironmak. Steelmak.,

2010, vol. 37, pp. 612–19.17. H. Jalkanen and L. Holappa: Treatise on Process Metallurgy, vol.

3, S. Seetharaman, Elsevier, Amsterdam, 2014, pp 223–70.18. V.D. Eisenhuttenleute and V.D.E.A.f.M. Grundlagen: Schlacke-

natlas., 2nd ed., Verlag Stahleisen, 1995.19. A. Smalcerz and R. Przylucki: Int. J. Thermophys., 2013, vol. 34,

pp. 667–79.20. W. Fiz, H. Heymann, and R. Heinke: J. Am. Ceram. Soc., 1969,

vol. 52, pp. 346–47.21. R. Inoue and H. Suito: ISIJ Int., 2006, vol. 46, pp. 174–79.22. K. Shimauchi, S. Kitamura, and H. Shibata: ISIJ Int., 2009,

vol. 49, pp. 505–11.23. F. Pahlevani, S. Kitamura, H. Shibata, and N. Maruoka: ISIJ Int.,

2010, vol. 50, pp. 822–29.24. M. Ek, J.C. Huber, G. Brosse, and D. Sichen: Ironmaking Steel-

making, 2013, vol. 40, pp. 305–11.

3170—VOLUME 49B, DECEMBER 2018 METALLURGICAL AND MATERIALS TRANSACTIONS B